Unfolding the medial temporal lobe to characterize neurodegeneration due to Alzheimer's disease pathology

Abstract

AbstractBackgroundMeasurements of medial temporal lobe (MTL) neurodegeneration derived using MRI have been shown to be sensitive to changes during the early stages of AD. The specificity of these measurements to tau neurofibrillary tangle (NFT) pathology is limited by other frequently comorbid non‐AD factors which also cause structural changes in the MTL. Here, we directly link changes in MTL structure to underlying NFT pathology by combining ex vivo MRI with ratings of NFT severity derived from serial histology using a dataset of 18 human MTL specimens. We hypothesize that such an analysis can be used to define MTL “hotspots” where in vivo measures will be more sensitive to disease progression in preclinical AD than current state of the art biomarkers.MethodEx vivo specimens from 18 donors were scanned at 0.2x0.2x0.2mm3 on 9.4T MRI. Following MRI scanning, the specimens underwent histological processing with staining for cytoarchitecture and in 15 specimens, immunohistochemistry (IHC) with the anti‐tau AT8 antibody. Using a topological unfolding method (DeKraker et al. 2018), we created 2D representations of the extra‐hippocampal cortex which implicitly align cortical folding patterns across specimens (Fig. 1). An average MTL subregion segmentation was generated in unfolded space using manual segmentations completed in 11 specimens on the basis of cytoarchitecture. Additionally, heat maps quantifying NFT burden in each of the specimens with anti‐tau IHC sections were generated using a deep learning algorithm (Yushkevich et al. 2021). Using the heatmaps and the average subregion segmentation, we investigated the relationship between NFT severity and cortical thickness.ResultCorrelation analysis between NFT measures and thickness (correcting for age) reveals strong associations in the entorhinal cortex and the border of Brodmann Area 35, consistent with the early Braak regions, and parts of Brodmann Area 36 (Fig. 2).ConclusionWe present an unfolding framework applied to the MTL cortex, which allows us to visualize, for the first time, the distribution of MTL subregions and NFT pathology in an unfolded space. This framework provides a promising tool for detailed investigation of structural changes due to NFT pathology while explicitly accounting for the complex topology of the MTL, thereby enhancing our understanding of early AD.